terraform covers state surgery, safe refactoring with moved/import/removed blocks, plan review, and drift control. addresses count vs for_each tradeoffs, lifecycle rules, module design, and common failure modes in infrastructure code.

fasting tracker logs intermittent and extended fasts with elapsed-time computation, stage estimation, and safety routing. scales break-fast guidance by fast duration and enforces electrolyte thresholds at 24h. red flags (hypoglycemia, orthostatic drop, disordered eating) stop protocol advice and escalate to clinician.

nicky-test enables agents to discover and execute cryptocurrency payments on behalf of users via the nicky platform, covering request lookup, asset selection, wallet generation, transaction reporting, and confirmation polling without requiring authentication.

typography covers measure, leading, tracking, weight, display vs text cuts, pairing, scale, rendering, and print mechanics with specific ratios, css properties, and failure signals tied to screen and print contexts.

scrapfly-alerting covers threshold-based monitoring on account metrics via rest api, mcp tools, or cli. grounds thresholds in baseline data and previews against history before committing, gated behind enterprise plan and feature flag.

accessibility covers wcag 2.1 aa compliance with semantic html, aria patterns, keyboard navigation, color contrast, and form accessibility. includes 12 preventable issue patterns and working code examples for dialogs and tabs.

agentnomos-governance-preflight enforces fail-closed evaluation of consequential agent actions before execution, checking actor identity, authority, scope, risk, and evidence through a six-step decision procedure that returns advisory allow, hold for review, or block with structured json output.

compound-eng-verification-before-completion enforces mandatory fresh evidence before any completion claim, with concrete gate functions, change-type-matched strategies, and adversarial probes to catch happy-path-only thinking.

happyhorse-1-0 wraps the runcomfy cli for text-to-video generation with native 1080p output, in-pass synchronized audio, and multi-shot character consistency across six prompt languages.

rhino3d-scripts covers authoring and debugging across three scripting surfaces (rhinopython, rhinoscript vbscript, and rhinocommon .net) for rhino 8+, with emphasis on geometry manipulation, command macros, and object selection workflows.

compound-eng-writing-tests covers test quality discipline across languages and frameworks, with emphasis on real assertions, edge case coverage, and anti-pattern recognition. guides test pyramid structure, mock boundaries, and bug-fix validation patterns.

claw-config provides agent-scoped, grounded configuration diagnosis and safe patching for openclaw agents. it grounds all field names and semantics in the installed binary schema and official docs, refusing hallucination or cross-agent writes. seven subcommands support read-only inventory, schema inspection, documentation lookup, dry-run planning, and atomic patching with automatic rollback.

firecrawl-api-al provides comprehensive instruction for scraping, crawling, mapping, and searching the web via firecrawl.dev. it covers operation selection, async crawl workflow, cost management, structured extraction, and source citation rules.

qqbot-send-local-media stages disposable copies of local files into a relay directory, sends via qqmedia tags, and cleans up only the staged copy while preserving originals. handles urls directly without staging.

scavio-instagram exposes 12 endpoints for reading instagram public data: profiles, posts, reels, stories, comments, followers, and search. costs range 2-10 credits per call; understand the identity quirks and field naming before chaining calls.

bunkerie seals encrypted identity and context checkpoints for ai agents, storing one offline snapshot controlled by a single key with no account recovery or key rotation.

scheduling-and-queue bridges finished social content to live posts via woopsocial, always confirming before publishing. it validates against platform rules, handles multi-account fanout, and degrades gracefully when disconnected.

publish-audit scans a skill folder against clawhub's release rules before upload, catching metadata mismatches, undeclared credentials, secrets, and file violations in a single pass with a clear verdict and fix list.

mysticx-tarot-drawer pulls tarot cards from a free, unauthenticated api supporting 13 spreads, 9 languages, and full 78-card browsing. no setup required, no rate limits.

privateclaw-plugin-setup covers installation, enablement, verification, and pairing of the PrivateClaw OpenClaw plugin, with preference for same-conversation QR delivery via /privateclaw commands and fallback to local CLI pairing.

graphql covers schema design, resolver debugging, and security hardening for multi-service graphs. addresses n+1 queries, null propagation, pagination, mutations, federation, and production deployment with situation-mapped procedures and configurable conventions.

collaborate structures two-person exchanges to test assumptions, resolve stuck decisions, and audit audience fit. it routes by undo cost and acceptance criteria clarity, caps exchanges at three rounds with movement tracking, and converges to one whole design plus a recorded objection.

z-zero-payments enables agents to purchase goods via gasless usdc on base or single-use virtual cards, with card data isolated from model context and explicit human approval required before any transaction executes.

agent-desktop is a cli tool for ai agents to observe and control desktop applications via native os accessibility trees. it outputs structured json with ref-based element identifiers, supports progressive skeleton traversal to reduce token consumption, and covers 54 commands across observation, interaction, keyboard/mouse, app lifecycle, and system operations.

seo-audit runs 12 hard-gate checks (h1, viewport, landmarks, title, canonical, og, images, csp, page weight, json-ld, url hygiene) over built static html via two zero-dependency node scripts, then reports ranked findings with exact fix references and an lcp playbook for core web vitals optimization.

paper-fetch resolves papers by doi or title and downloads pdfs from multiple sources in priority order: unpaywall, semantic scholar, arxiv, pubmed central, biorxiv/medrxiv, publisher direct (institutional), and sci-hub. returns structured json with full metadata, sources tried, and retry hints.

sol-dapp-builder guides solana product teams through architecture discovery, on-chain/off-chain boundaries, and production scaffolding. covers account design, authority models, stack choices, and deployment checklists with security and cost tradeoffs.

bookforge-scratch-refactoring-for-code-understanding teaches developers to restructure unfamiliar code freely on a disposable branch to build mental models, then discard the changes while preserving learning notes. targets legacy code comprehension before safe modification.

picsee-short-link exposes 15 PicSee tools via MCP for URL shortening, QR generation, and granular click analytics (daily, platform, referrer, region, audience). anonymous mode supports basic link creation; OAuth 2.1 unlocks full account management and attribution tracking.

quickbooks-online wraps the intuit api in a cli with 164 commands across 29 groups. covers ar, ap, chart of accounts, banking, reporting, and month-end workflows. all output is json by default for agent consumption.

hugging-face-api enables discovery of open-source models and datasets via free hub search, then routes chat and embedding inference through an openai-compatible interface with explicit cost controls and token security.

worldbank-al retrieves annual development, macroeconomic, and demographic indicators from world bank open data. supports cross-country comparisons, time-series retrieval, and requires no api key. integrates with worldbank-mcp server over stdio.

notion-second-brain routes telegram messages into a notion workspace, sorting books into a reading list and everything else into an inbox. it enforces workspace structure upfront and never asks clarifying questions.

claim-to-source-auditor-skill extracts verifiable claims from articles and reports, traces them to primary or reliable secondary sources, assigns pass/partial/missing/wrong/judgment verdicts with p0-p2 severity, and produces structured audit reports with cross-platform version comparison and regression tracking.

nginx covers reverse proxy config, load balancing, ssl/tls termination, caching, and static serving with explicit debugging patterns for common errors like 502/504 timeouts, websocket failures, and header inheritance traps.

bash covers reliable script authoring with explicit treatment of quoting, error handling, traps, and portability constraints. addresses the most common failure modes (word splitting, silent failures, filename edge cases) with concrete patterns and guardrails.

institutional-13f-tracker surfaces sec 13f filings to show who owns a stock, how institutional positions shifted quarter-over-quarter, and which activist funds took new stakes, via read-only api calls to aggregate and per-manager holdings.

fedica skill composes and schedules cross-platform social posts (x, linkedin, threads, mastodon, bluesky) via agent-browser, with explicit handling of timezone conversion, character limits per platform, and shadow-dom date picker mechanics.

skvm-general drives the skvm CLI to profile models, AOT-compile skills, run ad-hoc tasks, execute benchmarks, and manage compiled proposals across heterogeneous LLM backends.

fact-checking-explainers enforces that every checkable claim in an explainer traces to a source, gets corrected, or gets cut before delivery. it runs at research-time and post-draft gates, with special handling for code-based claims checked against real implementations.

agentic-cli-coding wraps safe, validation-aware editing via the oce command, providing structured workflows for reading, searching, and modifying source code across 25+ languages with automatic rollback on syntax failure.

drivethru-graphic-artist composites decorations onto product photos using deterministic ratio-based placement rules, detects garment bounding boxes, and self-reviews output with up to 3 correction passes before returning the mockup.

quakeproof wraps the usgs earthquake catalog to retrieve magnitude, distance, felt reports, and shakemap availability for a specific u.s. street address. designed for insurance claims, fema applications, and property due diligence, not earthquake news or predictions.

mask-driven image inpainting on runcomfy via cli, routing to z-image turbo for precise region edits (object removal, watermark cleanup) and description-based models when masks unavailable. covers strength tuning, prompting heuristics, and fallback workflows.

okx-a2a-payment wraps onchainos agent-to-agent payment operations: seller-side link creation with configurable expiry, buyer-side EIP-3009 signing and auto-polling to terminal state, and real-time status queries. explicitly delegates buyer-side trust verification to upstream callers.

add-literature searches the web for scholarly papers matching user keywords, extracts verified metadata from real sources, and publishes them to a shared platform with automatic deduplication by doi or url. optionally attaches open-access pdfs.